Artificial Intelligence in Managing Spasticity with Botulinum Toxin Type A-Insights from an Exploratory Pilot...

This study aimed to explore the potential role of artificial intelligence in optimizing botulinum toxin type A treatment for spasticity and to evaluate its alignment with expert clinical decisions. A comparative analysis was conducted using thirty hypothetical clinical cases involving individuals with spasticity resulting from various neurological conditions. Five rehabilitation physicians, each with more than five years of experience, participated in the study. An artificial intelligence model trained on scientific literature and clinical guidelines generated treatment recommendations, including target muscles and dosages, which were compared with those proposed independently by the physicians. The primary outcome was the level of agreement in muscle selection and dosage. The model demonstrated consistency and adherence to guidelines but showed limited adaptability in complex presentations, such as an adducted thigh and equinovarus foot. It generally recommended lower dosages and differed significantly from physicians in both muscle selection and treatment strategies. Artificial intelligence shows promise as a clinical support tool in spasticity management, offering standardized and reproducible recommendations. However, its limited capacity to interpret clinical subtleties currently restricts its practical application. Future models should integrate multimodal clinical data and real-time clinician feedback to better emulate expert decision-making processes.